Master Python Web Deployment: From CGI to uWSGI and Gunicorn

Developers familiar with PHP know that deploying a PHP site is straightforward, often requiring only a few file changes and supporting FastCGI out of the box. In contrast, Python web deployment is more complex due to a variety of tools and limited native server support.

CGI

Common Gateway Interface (CGI) defines a standard way for external programs to communicate with a web server. It creates a new process for each request, which leads to high overhead and poor performance under heavy load.

FastCGI

FastCGI improves on CGI by keeping interpreter processes alive, eliminating the need to fork for every request. This reduces CPU and memory usage and can increase performance by at least five times. FastCGI is language‑agnostic and supported by servers such as Apache, Nginx, and lighttpd.

WSGI

Python Web Server Gateway Interface (WSGI) is a low‑level standard that separates the web server (gateway) from the application or framework. It passes an environment dictionary and a callback function to the application, which returns the response. Middleware can modify requests, enable load balancing, or perform post‑processing.

uWSGI

uWSGI is a server that implements its own protocol, the uwsgi protocol, which is claimed to be ten times faster than FastCGI. It also supports WSGI, HTTP, and other protocols, offering high performance, low memory usage, multi‑app management, detailed logging, and extensive configurability.

Gunicorn

Gunicorn (Green Unicorn) is a pre‑forking WSGI server originally ported from the Ruby Unicorn project. It works well with Django, supports multiple worker types (sync, gevent, eventlet, etc.), and is easy to configure, though it lacks HTTP/1.1 support and has lower concurrency compared to uWSGI.

Tornado

Tornado is both an asynchronous Python framework and a non‑blocking HTTP server. When used as a standalone server it bypasses WSGI, but it can also run under WSGI via tornado.wsgi.WSGIContainer to integrate with other frameworks.

wsgiref

The built‑in wsgiref module provides a reference implementation of a WSGI server, used by Django’s development server and suitable for simple testing.

In practice, after developing a Django or Tornado application, you would deploy it with a production server such as uWSGI (often behind Nginx) or Gunicorn (also behind Nginx). Nginx handles static content and load balancing, while process managers like supervisord control the worker processes. Because of Python’s GIL, a multi‑process model is typically used, e.g., one master and two workers.

Source: titanjf, www.cnblogs.com/titanjf/p/python-web-deploy.html